Magnets called undulators stretch roughly 100 meters down a tunnel at SLAC National Accelerator Laboratory, with one side (right) producing hard x-rays and the other soft x-rays. (Credit: SLAC National Accelerator Laboratory)

Twelve years ago, physicists turned on the first x-ray laser, and since then it and several others around the world have proved themselves revolutionary probes of materials and molecules. But the devices, called x-ray free-electron lasers (XFELs), are only partially laserlike. In contrast to the pure, single-wavelength light emitted by conventional lasers, they produce noisy, chaotic beams. Now, physicists are developing a scheme that would enlist perfect diamond mirrors to make the x-ray pulses much more like ordinary laser beams and even more useful.

With two facilities now racing to stage proof-of-principle experiments as early as 2023, would-be users are taking notice. “I’m excited about the potential of this,” says Serena DeBeer, a chemist at the Max Planck Institute for Chemical Energy Conversion, who says the beams could be used to study the inner workings of enzymes as they catalyze reactions. But realizing such sophisticated XFELs may take 10 years and won’t be easy, warns Harald Sinn, an x-ray physicist at the European XFEL: “There are nightmares ahead.”

A conventional laser consists of a light-emitting material sitting between two mirrors. The carefully spaced mirrors form a cavity that resonates with light of the desired wavelength, just as an organ pipe rings with sound of a specific pitch. As the light passes back and forth through the material, it stimulates the stuff to produce more photons of the same wavelength, amplifying the light until a wave of identical photons marching in quantum mechanical lockstep, i.e. a laser beam, shines through one mirror, which is purposefully made imperfectly reflective.

This scheme won’t work for x-rays. Physicists lack both an obvious radiating material and, until recently, mirrors that will reflect x-rays at large enough angles to form a resonating cavity. So, they use a particle accelerator to fire a bunch of electrons down a vacuum pipe and through long trains of magnets called undulators, which shake the electrons side-to-side so they radiate x-ray photons. The light then travels along with the electrons and pushes them into microbunches, which wiggle in unison and radiate far more strongly, producing a burst of x-rays just femtoseconds long.

The first free-electrion laser flicked on in the 1970s, producing much longer wavelength microwaves. It was not until 2009 that physicists at SLAC National Accelerator Laboratory achieved the feat for “hard” x-rays, when they used the lab’s 3-kilometer-long linear accelerator to fire up the world’s first XFEL, the Linac Coherent Light Source (LCLS). Other countries have since built a half-dozen XFELS.

As with ordinary laser beams, x-rays from an XFEL arrive in smooth fronts, like ocean waves across a beach. A single XFEL pulse can scatter off a nanometer-size crystal and reveal its atomic structure, even as it blows the crystal to bits. Biologists have used XFELs to determine the structures of myriad proteins and other molecules that won’t form crystals big enough to be studied at less intense x-ray sources. But because an XFEL uses fluctuations in the density of the electron beam to begin to generate x-rays, one pulse varies from another in intensity, and each pulse has a wide and randomly distributed spectrum of wavelengths.

To squelch such noise, physicists have turned to an idea kicked around for decades, says Kwang-Je Kim, an accelerator physicist at Argonne National Laboratory. “People talked about it from time to time over drinks, but it was party conversation,” he says. “Nobody did any serious calculations” until the late 2000s, when Kim and others tackled the issue.

In an x-ray free-electron laser (XFEL), an undulator magnet shakes a bunch of energetic electrons sideways so they emit x-rays. The x-rays push the electrons into subbunches that, radiating in concert, then generate a noisy tsunami of x-rays. A twist on the concept could produce more consistent, smoother x-ray pulses.

The idea is to extract part of the x-ray pulse generated by one bunch of electrons and feed it back to the entrance of the undulators just in time to overlap with the next bunch of electrons. The recirculated x-rays would serve as a seed that causes the electrons to radiate more predictably. In repeated cycles, the x-ray pulses should become very pure and smooth, with a spread of wavelengths only 1/1000th as wide as ordinary XFEL pulses.

The plan requires very special mirrors, however. X-rays blast through most material, but for 100 years, physicists have known that a perfect crystal should reflect x-rays at certain angles, depending on the x-rays’ energy and the crystal’s structure and orientation, as the x-rays diffract off parallel planes of atoms in the crystal. The crystal also acts as a filter, as it reflects x-rays in a narrow range of wavelengths. Such crystal mirrors remained an aspiration until 2010, when Yuri Shvyd’ko, an x-ray physicist at Argonne, and colleagues showed small synthetic diamonds can reflect x-rays with 99% efficiency. Fortunately, an XFEL’s beam is less than 100 micrometers wide. “You don’t need a large crystal,” Shvyd’ko says. “You need a perfect crystal of small size.”

The scheme also requires a linear accelerator with a high repetition rate, to ensure the x-ray beam encounters a fresh bunch of electrons each time it rounds its circuit of mirrors. SLAC’s original accelerator is way too slow, firing 120 times a second. The European XFEL runs at 2.2 million cycles a second, so a cavity just 136 meters long would synchronize the x-rays with the electron bunches. SLAC is installing an accelerator that will run at 1 million cycles per second starting in 2022.

To test the essential elements for a cavity-based XFEL, physicists from Argonne, SLAC, and the Japanese lab Spring-8 plan to use four crystal mirrors to build a 66-meter-long cavity around seven LCLS undulators. By fiddling with SLAC’s current accelerator, they will shoot two bunches of electrons separated by 220 nanoseconds through the undulators and hope to show that recirculating x-rays from the first bunch make the second bunch radiate more efficiently. The system should be up and running in 2023, says Gabriel Marcus, an accelerator physicist at SLAC. Researchers at the European XFEL plan to implement a slightly different design by 2024. They hope to send up to 2700 electron bunches through the undulators and watch the laser beam grow stronger and smoother with each pass.

Patrick Rauer, an x-ray physicist at the University of Hamburg who has modeled the European XFEL project on a computer, cautions that the scheme will require extraordinary precision, with the millimeter-size diamonds aligned to a few millionths of a degree. “It’s a major problem,” Rauer says. “This is going to very difficult.” Ilya Agapov, an accelerator physicist at the German Electron Synchrotron laboratory, says that even harder will be maintaining the alignment as circulating x-rays heat the mirrors.

Still, potential users foresee major benefits. For example, Christian Gutt of the University of Siegen has used the European XFEL to study how proteins in solution diffuse and cluster on time scales as short as nanoseconds by studying correlations in the patterns of x-rays diffracted by the proteins. Those patterns would be far sharper with a cavity-based XFEL, he says. “That would be a game changer for us.”

With its extremely narrow spectrum, a cavity-based XFEL might even serve to control the quantum states of atomic nuclei much as atomic physicists now control the states of atoms with visible light, says Linda Young, an atomic physicist at Argonne. “It’s very wild,” she says. All it will take is a few mirrors—and a lot of hard work.